Centre for Economic Policy Research

ex-post yields. This certification effect was stronger for more junior securities and
less seasoned issues.
66
There also appears to be no support for the view that universal banks inflicted
low quality securities on the public. Kroszner and Rajan (1994) collected a sample
of industrial bonds underwritten by the securities affiliates of commercial banks
and by independent investment banks for 1921-29. Measured by Moody’s
Manuals and Poor’s Manuals ratings, the quality of bonds fell over time for both
classes of underwriters as the stock market boom proceeded, although the affili-
ates originated higher quality bonds. In tests of matched industrial bonds and
logit regressions, Kroszner and Rajan found that there were fewer defaults among
the affiliate-originated issues, measured in number and dollar value. Furthermore,
the losses for defaulted bonds originated by universal banks were no different
than the losses sustained by defaulted bonds issued by independent investment
banks.
67
Examining the cumulative default rates of universal bank and investment
bank underwritten securities issued between 1927 and 1929 with a probit model,
Puri (1994) found that issues underwritten by universal banks issues defaulted less 
frequently. Ang and Richardson (1994) similarly found that bonds underwritten
by bank affiliates had lower default rates. Thus, the public’s faith and the rating 
agencies’ judgements were upheld by experience. 
Security affiliates and investment banks did not underwrite the same mix of
securities, however. It appears that affiliates shied away from underwriting 
66 Conflicts of Interest in the Financial Services Industry

---

smaller, more junior securities and concentrated on underwriting larger older
firms with an emphasis on senior securities. Kroszner and Rajan (1994) argued
that investors were aware of the potential conflicts of interest but could not 
easily see behind the scenes to judge whether the banks were exploiting them.
While investors did not have the information that would have allowed them to
discriminate between issues that were tainted and untainted by conflicts of 
interest, they responded by applying a ‘lemons-market’ discount to these bonds.
Evidence for this problem is suggested by the fact that ratings were good predic-
tors of default for investment bank underwritten issues but poor predictors for
affiliate underwritten issues. Universal banks compensated by underwriting 
higher quality securities, which were less information sensitive but for which they
still enjoyed advantages in information collection.
What is even more striking is that the commercial banks engaged in investment
banking recognized that they could reassure the public of the value of their 
certification by using a securities affiliate. Initially most banks in the United States
in the 1920s underwrote bonds through a securities department inside the bank,
like a German universal bank. In addition to allowing them to issue equities, 
however, banks soon appear to have found that they could mitigate conflicts by
moving underwriting to an affiliate, thus distancing this activity from lending
and gaining credibility for their certification. By creating a securities affiliate, the
managerial hierarchies could be separated and questions of information and 
coordination would be handled by top management instead of at the department
level. The incentives and compensation of the managers of the affiliate could be
quite distinct from those of the loan officers. The balance sheets, income 
statements and accounting information for the parent banks and their separate
affiliates provided greater transparency.
Kroszner and Rajan (1997) found strong evidence for the ability of commercial
banks to signal their control of conflicts through their organizational structure.
Drawing on all new public securities issued for the period 1925-29, they examined
906 securities underwritten by 43 internal bond departments and 32 securities
affiliates. Adjusting for characteristics of the bonds and the banks, the difference
in the yield to maturity at the offering date for the bonds was 14 to 23 basis points
lower. The market seemed alert to even more subtle indications of separation.
Affiliates whose board of directors exactly matched the parent bank’s board of
directors had the issues that they underwrote discounted by approximately the
same degree as those of internal bond departments. Furthermore, securities
departments and affiliates underwrote different mixes of securities. Compared to
internal departments, affiliates tended to underwrite more junior securities and
issues for smaller, younger and more indebted firms. These attributes suggest that
affiliates had more credibility to certify these more information intensive 
securities.
Contrary to the impression in the Congressional hearings, the potential for
conflicts of interest was not something discovered the day after the 1929 crash of
the stock market. Some bankers were well aware of the problem even in the 
middle of the decade, before the market heated up. The Farmers’ Loan and Trust
Company of New York told the Commercial and Financial Chronicle in 1925 that:
Over time, managers became aware that the prices of issues underwritten by
Conflicts of Interest in Universal Banking 67
‘Due to our policy and firm conviction that, as a trustee, we should never
place ourselves in the position of a buyer and seller of securities at the same
time, we have never had a bond department. Our whole security depart-
ment is organized for the impartial study of securities for the benefit of our
customers and not for the sale of bonds to the public’ (quoted in Kroszner
and Rajan, 1997).

---

internal departments could be increased by switching to the more credible form
of a securities affiliate. Not only did new entrants choose the affiliate form to
enter the securities business, but many bond departments were also converted to
affiliates. While the two organizational types divided the number of new issues
almost exactly in half in 1925, separate affiliates garnered 82% of all issues by
1929 (Kroszner and Rajan, 1997).
68
Thus, recent scholarship has almost 
completely overturned this conventional wisdom about universal banking in the
critical era of the 1920s.